×

签到

分享到微信

打开微信,使用扫一扫进入页面后,点击右上角菜单,

点击“发送给朋友”或“分享到朋友圈”完成分享

在进行tensorflow集成时报错 zrr19992020-07-01 18:31:39 回复 2 查看
在进行tensorflow集成时报错
分享到:

tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc:1924:96: error: cannot convert 'MLUTensor** {aka cnmlTensor**}' to 'cnmlPluginPowerDifferenceOpParam_t {aka cnmlPluginPowerDifferenceOpParam*}' for argument '2' to 'cnmlStatus_t cnmlCreatePluginPowerDifferenceOp(cnml Op**, cnmlPluginPowerDifferenceOpParam_t, cnmlTensor**, cnmlTensor**)'
   CNML_RETURN_STATUS(cnmlCreatePluginPowerDifferenceOp(op, inputs_ptr, input3, outputs_ptr, len));
                                                                                                ^
./tensorflow/stream_executor/mlu/mlu_api/lib_ops/util.h:135:29: note: in definition of macro 'TF_CNML_CHECK'
     MLUStatus mlu_status = (__VA_ARGS__);              \
                             ^
tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc:1924:3: note: in expansion of macro 'CNML_RETURN_STATUS'
   CNML_RETURN_STATUS(cnmlCreatePluginPowerDifferenceOp(op, inputs_ptr, input3, outputs_ptr, len));
   ^
tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc: In function 'tensorflow::Status stream_executor::mlu::lib::ComputePowerDifferenceOp(MLU Op*, MLUCnrtQueue*, void*, void*, void*)':
tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc:1933:41: error: invalid conversion from 'void**' to 'int' [-fpermissive]
       op, inputs_ptr, outputs_ptr, queue));
                                         ^
./tensorflow/stream_executor/mlu/mlu_api/lib_ops/util.h:135:29: note: in definition of macro 'TF_CNML_CHECK'
     MLUStatus mlu_status = (__VA_ARGS__);              \
                             ^
tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc:1932:3: note: in expansion of macro 'CNML_RETURN_STATUS'
   CNML_RETURN_STATUS(cnmlComputePluginPowerDifferenceOpForward(
   ^
tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc:1933:41: error: cannot convert 'MLUCnrtQueue* {aka cnrtQueue*}' to 'void**' for argument '4' to 'cnmlStatus_t cnmlComputePluginPowerDifferenceOpForward(cnml Op_t, void**, int, void**, int, cnrtInvokeFuncParam_t*, cnrtQueue_t)'
       op, inputs_ptr, outputs_ptr, queue));
                                         ^
./tensorflow/stream_executor/mlu/mlu_api/lib_ops/util.h:135:29: note: in definition of macro 'TF_CNML_CHECK'
     MLUStatus mlu_status = (__VA_ARGS__);              \
                             ^
tensorflow/stream_executor/mlu/mlu_api/lib_ops/mlu_lib_ops.cc:1932:3: note: in expansion of macro 'CNML_RETURN_STATUS'
   CNML_RETURN_STATUS(cnmlComputePluginPowerDifferenceOpForward(
   ^
Target //tensorflow/tools/pip_package:build_pip_package failed to build
INFO: Elapsed time: 68.704s, Critical Path: 16.88s
INFO: 180 processes: 180 local.
FAILED: Build did NOT complete successfully

image.png

版权所有 © 2024 寒武纪 Cambricon.com 备案/许可证号:京ICP备17003415号-1
关闭